Integrated method and apparatus for character and symbol recognition

Abstract

An apparatus and method for recognizing written items including characters and symbols is disclosed. A scanner is utilized for scanning the characters and symbols. A recognition module is used to confidently identify the scanned characters and symbols. The recognition module includes a mechanism for recognizing characters. This mechanism includes a character recognition rule base which yields identified characters and unidentified characters. The unidentified characters are conveyed to neural networks for recognition. The recognition module also includes a mechanism for recognizing symbols. This mechanism includes a symbol recognition rule base which yields identified symbols and unidentified symbols. The unidentified symbols are conveyed to neural networks for recognition. The recognition module also includes a mechanism for context processing of the characters and symbols. The context processor utilizes a multilevel blackboard with a number of ascending levels. The blackboard combines characters and symbols into logical units. The blackboard also serves to verify the identity of the characters and symbols. The resultant data from the recognition module is placed in a metafile which is conveyed to a modification module. The modification module includes graphics editing software to edit the scanned characters and symbols. After editing, a translator is used to convert the scanned characters and symbols to a format suitable for a CAD system or data base.

What is claimed is: 
     
       1. A method of recognizing written items including characters and symbols, comprising the steps of: feeding said items to rulebases for recognition;   processing selected features of items which are not recognized during said feeding step; and   conveying said selected features corresponding to said items which are not recognized during said feeding step to neural networks for recognition.   
     
     
       2. The method of claim 1 further comprising the step of: delivering said items to a context a blackboard with a plurality of ascending levels.   
     
     
       3. The method of claim 2 wherein said delivering step includes the steps of: (A) analyzing items on a level;   (B) harvesting items on said level after said analyzing step;   (C) distributing items on a higher level after said harvesting step; and   (D) repeating steps (A) through (C) for a plurality of ascending levels.   
     
     
       4. A method of recognizing written items including characters and symbols, comprising the steps of: (A) processing said characters, said character processing including: (1) recognizing fully formed characters by: (a) transmitting characters to a full character recognition rulebase, said full character recognition rulebase yielding identified characters and unidentified characters;   (b) conveying said unidentified characters to full character neural networks;     (2) isolating broken characters and recognizing them by: (a) conveying said broken characters to broken character neural networks;       (B) processing said symbols, said symbol processing including: (1) transmitting said symbols to a symbol recognition rulebase, said symbol recognition rulebase yielding identified symbols and unidentified symbols;   (2) conveying said unidentified symbols to symbol recognition neural networks; and     (C) context processing said characters and symbols, said context processing including the use of a multilevel blackboard with a plurality of ascending levels, said blackboard combining characters and symbols into logical units and further serving to verify the recognition of said characters and symbols.   
     
     
       5. The method of claim 4 wherein said character processing step further includes the steps of: vectorizing said unidentified characters to form vectorized characters; and   extracting features from said vectorized characters to form extracted features and feeding said extracted features to said full character neural networks.   
     
     
       6. The method of claim 5 wherein said extracted features include: the x-coordinate of the centroid of the end-points of all vectors making up the boundary of said unidentified character; and   the y-coordinate of the centroid of the end-points of all vectors making up the boundary of said unidentified character.   
     
     
       7. The method of claim 5 wherein said extracted features include: the total slantiness of said unidentified character;   the mean slope of said unidentified character; and   the total angle of said unidentified character.   
     
     
       8. The method of claim 4 wherein said character processing step further includes the step of: enhancing features of said broken characters prior to conveying said characters to said broken character neural networks, said enhancing step including the steps of scale rastering said broken characters and applying regionalized convolution masks to said broken characters.   
     
     
       9. The method of claim 4 wherein said symbol processing step further includes the steps of: vectorizing said symbols;   enhancing features of said symbols prior to conveying said symbols to said symbol recognition neural networks, said enhancing step including the steps of scale rastering said symbols and applying regionalized convolution masks to said symbols.   
     
     
       10. The method of claim 4 wherein said context processing step includes the steps of: (A) analyzing items on a level of said blackboard;   (B) harvesting items on said level after said analyzing step;   (C) distributing items on a higher level of said blackboard after said harvesting step; and   (D) repeating steps (A) through (C) for a plurality of ascending levels of said blackboard.   
     
     
       11. The method of claim 4 wherein said context processing is utilized to move said characters and said symbols through said ascending levels of said blackboard, said characters and said symbols moving through a connector level. 
     
     
       12. The method of claim 4 wherein said context processing is utilized to move said characters and said symbols through said ascending levels of said blackboard, said characters and said symbols moving through an associativity level. 
     
     
       13. The method of claim 4 wherein said context processing is utilized to move said characters and said symbols through said ascending levels of said blackboard, said characters and said symbols moving through a geometry level. 
     
     
       14. The method of claim 4 wherein said context processing is utilized to move said characters and said symbols through said ascending levels of said blackboard, said characters and said symbols moving through a polygon level. 
     
     
       15. The method of claim 4 wherein said context processing is utilized to move said characters and said symbols through said ascending levels of said blackboard, said characters and said symbols moving through an arrow level. 
     
     
       16. The method of claim 4 wherein said context processing is utilized to move said characters and said symbols through said ascending levels of said blackboard, said characters and said symbols moving through a text level. 
     
     
       17. The method of claim 4 wherein said context processing is utilized to move said characters and said symbols through said ascending levels of said blackboard, said characters and said symbols moving through a virtual geometry level. 
     
     
       18. The method of claim 4 wherein said context processing is utilized to move said characters and said symbols through said ascending levels of said blackboard, said characters and said symbols moving through a composite level. 
     
     
       19. The method of claim 4 wherein said context processing is utilized to move said characters and said symbols through said ascending levels of said blackboard, said characters and said symbols moving through a meta-symbol level. 
     
     
       20. The method of claim 4 wherein said context processing is utilized to move said characters and said symbols through said ascending levels of said blackboard, said characters and said symbols moving through a junk level. 
     
     
       21. A method of recognizing written items including characters and symbols, comprising the steps of: (A) classifying said items as either characters or symbols;   (B) processing said characters, said character processing including: (1) recognizing fully formed characters by: (a) transmitting said characters to a full character recognition rulebase, said full character recognition rulebase yielding identified characters and unidentified characters;   (b) vectorizing said unidentified characters to form vectorized characters;   (c) extracting features from said vectorized characters to form extracted features;   (d) conveying said extracted features to full character neural networks;     (2) isolating broken characters; (a) enhancing the features of said broken characters to form enhanced broken characters;   (b) conveying said enhanced broken characters to broken character neural networks;       (C) processing said symbols, said symbol processing including: (1) transmitting said symbols to a symbol recognition rulebase, said symbol recognition rulebase yielding identified symbols and unidentified symbols;   (2) enhancing the features of said unidentified symbols to form enhanced symbols;   (3) conveying said enhanced symbols to symbol recognition neural networks; and     (D) context processing said characters and symbols, said context processing including the use of a multilevel blackboard with a plurality of ascending levels, said blackboard combining characters and symbols into logical units and further serving to verify the recognition of said characters and symbols.   
     
     
       22. The method of claim 21 wherein said extracted features include: the x-coordinate of the centroid of the end-points of all vectors making up the boundary of said unidentified character; and   the y-coordinate of the centroid of the end-points of all vectors making up the boundary of said unidentified character.   
     
     
       23. The method of claim 21 wherein said extracted features include: the total slantiness of said unidentified character;   the mean slope of said unidentified character; and   the total angle of said unidentified character.   
     
     
       24. The method of claim 21 wherein said broken character enhancing step includes the steps of scale rastering said broken characters and applying regionalized convolution masks to said broken characters. 
     
     
       25. The method of claim 21 wherein said symbol enhancing step includes the steps of scale rastering said symbols and applying regionalized convolution masks to said symbols. 
     
     
       26. The method of claim 21 wherein said context processing step includes the steps of: (A) analyzing items on a level of said blackboard;   (B) harvesting items on said level after said analyzing step;   (C) distributing items on a higher level of said blackboard after said harvesting step; and   (D) repeating steps (A) through (C) for a plurality of ascending levels of said blackboard.   
     
     
       27. The method of claim 26 wherein said context processing is utilized to move said items through said ascending levels of said blackboard, said context processing moving said items through a connector level and an associativity level. 
     
     
       28. The method of claim 26 wherein said context processing is utilized to move said items through said ascending levels of said blackboard, said context processing moving said items through a geometry level, an arrow level, a text level, and a meta-symbol level. 
     
     
       29. The method of claim 26 wherein said context processing is utilized to move said items through said ascending levels of said blackboard, said context processing moving said items through a polygon level, a virtual geometry level, and a composite level. 
     
     
       30. An apparatus for recognizing written items including characters and symbols, comprising: (A) a scanner for scanning said characters and symbols;   (B) a recognition module, said recognition module including: (1) means for processing said characters, said character processing means including: (a) means for recognizing fully formed characters by: (i) means for transmitting said fully formed characters to a full character recognition rulebase, said full character recognition rulebase yielding identified characters and unidentified characters;   (ii) means for conveying said unidentified characters to full character neural networks;     (b) means for isolating and recognizing broken characters, said means including: (i) means for conveying said broken characters to broken character neural networks;       (2) means for processing said symbols, said symbol processing means including: (a) means for transmitting said symbols to a symbol recognition rulebase, said symbol recognition rulebase yielding identified symbols and unidentified symbols;   (b) means for conveying said unidentified symbols to symbol recognition neural networks;     (3) means for context processing said characters and symbols, said context processing including the use of a multilevel blackboard with a plurality of ascending levels, said blackboard combining characters and symbols into logical units and further serving to verify the recognition of said characters and symbols;     (C) a modification module including graphics editing software to edit said scanned characters and symbols; and   (D) a translator to convert said scanned characters and symbols to a format suitable for a CAD system or data base.   
     
     
       31. The apparatus of claim 30 wherein said character processing means includes: means for vectorizing said unidentified characters to form vectorized characters; and   means for extracting features from said vectorized characters to form extracted features and means for feeding said extracted features to said full character neural networks.   
     
     
       32. The apparatus of claim 30 wherein said character processing means further include: means for enhancing features of said broken characters prior to conveying said characters to said broken character neural networks, said enhancing means including means for scale rastering said broken character and means for applying regionalized convolution masks to said broken character.   
     
     
       33. The apparatus of claim 30 wherein said symbol processing means further includes: means for vectorizing said symbols; and   means for enhancing features of said symbols prior to conveying said symbols to said symbol recognition neural networks, said enhancing means including means for scale rastering said symbol and means for applying regionalized convolution masks to said symbol.   
     
     
       34. The apparatus of claim 30 wherein said means for context processing includes: (A) means for analyzing items on a level of said blackboard;   (B) means for harvesting items on said level;   (C) means for distributing items on a higher level of said blackboard after said harvesting step; and   (D) means for invoking said analyzing means, said harvesting means, and said distributing means for a plurality of ascending levels of said blackboard.   
     
     
       35. The apparatus of claim 34 wherein said context processing means is utilized to move said characters and said symbols through said ascending levels of said blackboard, said characters and said symbols moving through a connector level and an associativity level. 
     
     
       36. The apparatus of claim 34 wherein said context processing means is utilized to move said characters and said symbols through said ascending levels of said blackboard, said characters and said symbols moving through a geometry level, an arrow level, a text level, and a metasymbol level. 
     
     
       37. The apparatus of claim 34 wherein said context processing means is utilized to move said characters and said symbols through said ascending levels of said blackboard, said characters and said symbols moving through a polygon level, a virtual geometry level, and a composite level.